“A good minimization of the empirical risk depends in many respects on the art of the researcher.” Vladimir N. Vapnik ( 1999 )